Energy storage configuration method, system and equipment and computer readable storage medium

Abstract

The invention discloses an energy storage configuration method, system and device and a computer readable storage medium, and relates to the technical field of energy storage. Performing feature extraction on the historical operation data according to the first time granularity to obtain a random scene set; determining a first decision variable of energy storage at the first time granularity; according to the first decision variable, generating a first optimization objective function of the stored energy at the first time granularity; generating a first constraint condition of the stored energy in the first time granularity; based on the random scene set, the first optimization objective function and the first constraint condition are processed through a Benders decomposition algorithm, and first configuration information of energy storage at the first time granularity is generated; based on the first configuration information, generating second configuration information of energy storage at a second time granularity; and based on the second configuration information, generating third configuration information of the energy storage at the third time granularity. According to the invention, the reliability, adaptability and robustness of energy storage configuration are improved.

Description

Technical Field

[0001] This application relates to the field of energy storage technology, and more specifically, to an energy storage configuration method, system, device, and computer-readable storage medium. Background Technology

[0002] Currently, energy storage needs to be configured according to requirements during its use. For example, a multi-objective stochastic optimization configuration method can be used for energy storage configuration. This scheme uses the k-means algorithm to extract monthly characteristic information of renewable energy power generation, forms a typical daily dataset, establishes an optimal power flow model, and solves the energy storage configuration based on second-order cone optimization.

[0003] However, this method mainly focuses on the randomness of renewable energy and reduces data dimensionality through clustering, but it does not fully consider the random characteristics of multiple time scales and their coupling effects, resulting in poor power supply reliability of energy storage and poor resistance to multivariate random disturbances.

[0004] In summary, improving the reliability, adaptability, and robustness of energy storage configurations is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] The purpose of this application is to provide an energy storage configuration method that can, to a certain extent, solve the technical problems of how to improve the reliability, adaptability, and robustness of energy storage configuration. This application also provides an energy storage configuration system, electronic equipment, and a computer-readable storage medium.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] An energy storage configuration method, comprising:

[0008] Obtain historical operational data of energy storage;

[0009] Feature extraction is performed on the historical operational data according to the first time granularity to obtain a set of random scenarios with coupled random variable statistical features, temporal correlation and spatial coupling;

[0010] Determine the first decision variable for energy storage at the first time granularity, the first decision variable including energy storage configuration capacity, energy storage configuration power and energy storage type selection;

[0011] Based on the first decision variable, generate a first optimization objective function for energy storage at the first time granularity;

[0012] Generate the first constraint condition for energy storage at the first time granularity;

[0013] Based on the random scene set, the first optimization objective function and the first constraint are processed by the Benders decomposition algorithm to generate first configuration information for energy storage at the first time granularity;

[0014] Based on the first configuration information, second configuration information for energy storage is generated at a second time granularity.

[0015] Based on the second configuration information, third configuration information for energy storage at a third time granularity is generated;

[0016] Wherein, the first time granularity is greater than the second time granularity, and the second time granularity is greater than the third time granularity.

[0017] Preferably, the first decision variable for determining energy storage at the first time granularity includes:

[0018] Determine a one-stage variable for energy storage at the first time granularity, wherein the one-stage variable includes energy storage configuration capacity, energy storage configuration power, and energy storage type selection;

[0019] Determine the two-stage variables for energy storage at the first time granularity, wherein the two-stage variables include the moment-to-moment charging power, moment-to-moment discharging power, and moment-to-moment state of charge of the random scenario set.

[0020] The first-stage variables and the second-stage variables are used as the first decision variables.

[0021] Preferably, the step of processing the first optimization objective function and the first constraint condition using the Benders decomposition algorithm based on the random scene set to generate first configuration information for energy storage at the first time granularity includes:

[0022] The random scene set is reduced in number by synchronous back-substitution elimination or K-means clustering to obtain the target scene set;

[0023] Based on the target scenario set, the first optimization objective function and the first constraint are processed by the Benders decomposition algorithm to generate first configuration information for energy storage at the first time granularity.

[0024] Preferably, generating second configuration information for energy storage at a second time granularity based on the first configuration information includes:

[0025] Determine the uncertainty vector for processing energy storage at the second time granularity;

[0026] The historical operational data is collected according to the uncertainty vector to obtain a sample set;

[0027] Based on the sample set, construct a moment uncertainty set based on one-matrix and two-matrix uncertainties;

[0028] Based on the first configuration information and the moment uncertainty set, a two-stage sub-Bruvian optimization model for energy storage at the second time granularity is established.

[0029] The two-stage sub-Brubon optimization model is processed by a column and constraint generation algorithm to generate second configuration information for energy storage at the second time granularity.

[0030] Preferably, the step of establishing a two-stage sub-Brønsted optimization model for energy storage at the second time granularity based on the first configuration information and the moment uncertainty set includes:

[0031] Determine the second decision variable for energy storage at the second time granularity;

[0032] Based on the moment uncertainty set and the second decision variable, a second optimization objective function for energy storage at the second time granularity is generated;

[0033] Based on the first configuration information, a second constraint condition for energy storage at the second time granularity is generated;

[0034] The second optimization objective function and the second constraint are used as a two-stage sub-Brupbond optimization model for energy storage at the second time granularity.

[0035] Preferably, the step of generating third configuration information for energy storage at a third time granularity based on the second configuration information includes:

[0036] Determine the third decision variable for energy storage at the third time granularity;

[0037] Based on the third decision variable, a third optimization objective function for energy storage is generated at the third time granularity;

[0038] Based on the second configuration information, a third constraint condition for energy storage is generated at the second time granularity;

[0039] The third optimization objective function and the third constraint are processed by deep reinforcement learning methods to generate third configuration information for energy storage at the third time granularity.

[0040] Preferably, the step of processing the third optimization objective function and the third constraint condition using a deep reinforcement learning method to generate third configuration information for energy storage at the third time granularity includes:

[0041] The state vector is determined based on the energy storage frequency, state of charge, power, random disturbances, and periodic pattern learning time.

[0042] Determine the action vector based on the energy storage control method;

[0043] A reward function is generated based on the energy storage evaluation indicators;

[0044] Based on the state vector, the action vector, and the reward function, a deep reinforcement learning agent is determined;

[0045] The third optimization objective function and the third constraint are processed by a deep reinforcement learning agent to generate third configuration information for energy storage at the third time granularity.

[0046] An energy storage configuration system, comprising:

[0047] The historical operation data acquisition module is used to acquire historical operation data of energy storage.

[0048] The random scene set generation module is used to extract features from the historical running data according to the first time granularity to obtain a random scene set with coupled random variable statistical features, temporal correlation and spatial coupling.

[0049] The first decision variable determination module is used to determine the first decision variable of energy storage at the first time granularity. The first decision variable includes energy storage configuration capacity, energy storage configuration power and energy storage type selection.

[0050] The first optimization objective function generation module is used to generate a first optimization objective function for energy storage at the first time granularity based on the first decision variable.

[0051] The first constraint generation module is used to generate the first constraint for energy storage at the first time granularity.

[0052] The first configuration information generation module is used to process the first optimization objective function and the first constraint condition based on the random scene set using the Benders decomposition algorithm to generate first configuration information for energy storage at the first time granularity.

[0053] The second configuration information generation module is used to generate second configuration information for energy storage at a second time granularity based on the first configuration information.

[0054] The third configuration information generation module is used to generate third configuration information for energy storage at a third time granularity based on the second configuration information.

[0055] Wherein, the first time granularity is greater than the second time granularity, and the second time granularity is greater than the third time granularity.

[0056] An electronic device, comprising:

[0057] Memory, used to store computer programs;

[0058] A processor, used to execute the computer program to implement the steps of any of the above-described energy storage configuration methods.

[0059] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of any of the above-described energy storage configuration methods.

[0060] This application provides an energy storage configuration method, which involves acquiring historical operation data of energy storage; extracting features from the historical operation data according to a first time granularity to obtain a set of random scenarios with coupled random variable statistical characteristics, temporal correlation, and spatial coupling; determining a first decision variable for energy storage at the first time granularity, the first decision variable including energy storage configuration capacity, energy storage configuration power, and energy storage type selection; generating a first optimization objective function for energy storage at the first time granularity based on the first decision variable; generating a first constraint condition for energy storage at the first time granularity; processing the first optimization objective function and the first constraint condition using the Benders decomposition algorithm based on the random scenario set to generate first configuration information for energy storage at the first time granularity; generating second configuration information for energy storage at a second time granularity based on the first configuration information; and generating third configuration information for energy storage at a third time granularity based on the second configuration information; wherein the first time granularity is greater than the second time granularity, and the second time granularity is greater than the third time granularity. In this application, features are extracted from historical operational data at a first time granularity to obtain a set of random scenarios with coupled random variable statistical characteristics, temporal correlation, and spatial coupling. This achieves an accurate characterization of the spatiotemporal correlation characteristics of multivariate random disturbances. Then, based on the first decision variable, the first optimization objective function, and the first constraints, an optimization architecture for energy storage at the first time granularity is constructed. Based on the set of random scenarios, the first configuration information for energy storage at the first time granularity is generated using the Benders decomposition algorithm, significantly improving the adaptability and robustness of the energy storage configuration scheme in complex real-world environments. Subsequently, a hierarchical planning framework is constructed through second and third time granularities, enabling energy storage configuration according to different time granularities. This optimizes long-term energy storage configuration and short-term operation, overcomes the limitations of traditional single-time-scale optimization, and improves the reliability of energy storage configuration. The energy storage configuration system, electronic device, and computer-readable storage medium provided in this application also solve corresponding technical problems. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0062] Figure 1 A flowchart illustrating an energy storage configuration method provided in this application embodiment;

[0063] Figure 2 This is a data processing diagram of the energy storage configuration scheme in this application;

[0064] Figure 3 This is a schematic diagram of the structure of an energy storage configuration system provided in an embodiment of this application;

[0065] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application;

[0066] Figure 5 This is another structural schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0067] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0068] Please see Figure 1 , Figure 1 This is a flowchart of an energy storage configuration method provided in an embodiment of this application.

[0069] An energy storage configuration method provided in this application embodiment may include the following steps:

[0070] Step S101: Obtain historical operating data of energy storage.

[0071] In practical applications, historical operating data of energy storage can be acquired first, so that the energy storage can be configured subsequently using this historical data. The type of historical operating data can be flexibly determined according to the application scenario, and may include historical wind and solar power output, load data, etc.

[0072] Step S102: Extract features from historical running data according to the first time granularity to obtain a set of random scenarios with coupled random variable statistical features, temporal correlation and spatial coupling.

[0073] In practical applications, existing methods typically handle various random factors individually when configuring energy storage, neglecting the coupling relationships between multiple random disturbances. For example, they ignore the correlation between tidal energy and photovoltaic output, and the spatiotemporal correlation between electric vehicle charging demand and grid load. This independent approach leads to poor performance of energy storage configuration schemes in complex real-world environments. To address this issue, this application requires feature extraction from historical operating data at a first time granularity to obtain a set of random scenarios with coupled random variable statistical characteristics, temporal correlations, and spatial coupling. The first time granularity can be year or month, etc.

[0074] In an exemplary embodiment, a random scene set can be generated based on the coupled scenarios of Copula. For example, for each month's historical operating data, a time-varying Copula function can be used to model the nonlinear dependency relationship between "wind power-solar power-load" within that month, such as the deterministic relationship of "high wind power at night, zero solar power" and the statistical correlation of "weak wind power on cloudy days". Then, the monthly scenes generated for 12 months are reasonably spliced ​​according to the transitional laws of climatology to form a scene s representing a complete year. This process is repeated to generate tens of thousands (S) different annual scenes {ξ1,ξ2,...,ξ... S As a random set of scenarios, each scenario contains data on 8760 hours of wind and solar load. The annual set of scenarios Ω is the basis for calculating the "expected cost" in subsequent optimizations.

[0075] Step S103: Determine the first decision variable for energy storage at the first time granularity. The first decision variable includes the energy storage configuration capacity, energy storage configuration power, and energy storage type selection.

[0076] In practical applications, it is necessary to determine the first decision variable for energy storage at the first time granularity. The first decision variable includes the energy storage configuration capacity, energy storage configuration power, and energy storage type selection, so that the first decision variable can be used to solve the first configuration information of energy storage at the first time granularity.

[0077] In an exemplary embodiment, during the process of determining the first decision variable of energy storage at the first time granularity, a first-stage variable of energy storage at the first time granularity can be determined, the first-stage variable including the energy storage configuration capacity ( (unit: MWh), energy storage configuration power ( (unit: MW) and selection of energy storage type ( 0-1 variables); determine the two-stage variables of energy storage at the first time granularity, the two-stage variables include the moment-to-moment charging power of the random scenario set ( ), discharge power at any time ( ) and the state of charge at any time ( The first-stage and second-stage variables are used as the first decision variables.

[0078] Step S104: Based on the first decision variable, generate the first optimization objective function for energy storage at the first time granularity.

[0079] In practical applications, it is necessary to generate a first optimization objective function for energy storage at the first time granularity based on the first decision variable, so that the first optimization objective function can be used to generate the first configuration information of energy storage at the first time granularity.

[0080] In an exemplary embodiment, the first optimization objective function can be: , This represents the annual investment cost, which is calculated by discounting the initial investment to each year using the depreciation rate. This represents an approximation of the expected operating cost, obtained by averaging the operating costs of all S scenarios. The operating cost includes electricity purchase cost, equipment depreciation, penalty cost, etc. The wind and solar load data for scene s at time t; It represents the set of charging power, discharging power, and state of charge at any given time.

[0081] Step S105: Generate the first constraint condition for energy storage at the first time granularity.

[0082] In practical applications, it is also necessary to generate the first constraint conditions for energy storage at the first time granularity based on the configuration requirements of energy storage, so that the first constraint conditions can be used to generate the first configuration information of energy storage at the first time granularity.

[0083] In an exemplary embodiment, the first constraint can be flexibly determined according to the application scenario, and may include one-stage constraints and two-stage constraints. One-stage constraints include investment budget constraints, installation location constraints, and technology matching constraints. Investment budget constraints are used to ensure that the total investment does not exceed the available budget. Installation location constraints are used to constrain space availability and grid connection conditions. Space availability includes floor space, and grid connection conditions include voltage level compatibility, short-circuit capacity meeting stability requirements, and line capacity ensuring no overload occurs after energy storage configuration. Technology matching constraints are used to ensure the complementarity of power-energy characteristics between different energy storage technologies. This mainly considers the balance issues that need to be considered when combining power-type and energy-type energy storage. In the specific coordinated scheduling process, filters can be used to achieve this. For example, high-frequency fluctuation components can be allocated to power-type technologies (those with the fastest response), while low-frequency components can be allocated to energy-type technologies, satisfying their respective technical limitations (capacity, power). Simultaneously, the optimization between the two aims to minimize cycle cost, selecting the most economically efficient allocation mode possible. The two-stage constraints require that each scenario s must satisfy: power balance, i.e., PV_s + wind_s + discharge_s - charge_s = load_s + external transmission_s; energy storage operation, i.e., SOC_s(t+1) = SOC_s(t) + η_char * charge_s(t) - discharge_s(t) / η_dis; and energy storage physical constraints, i.e., 0 ≤ charge_s(t) ≤ x. pow , 0≤SOC_s(t)≤xcap; power grid safety, that is, line transmission power ≤ limit.

[0084] Step S106: Based on the random scene set, the first optimization objective function and the first constraint condition are processed by the Benders decomposition algorithm to generate the first configuration information of energy storage at the first time granularity.

[0085] In practical applications, existing stochastic optimization models often face difficulties in solving problems. For example, non-convex nonlinear constraints and high-dimensional binary variables (such as energy storage start-up and shutdown and network topology) greatly increase the complexity of the problem. Although some studies have adopted linearization and convex relaxation techniques, they often sacrifice model accuracy, and the complex state-of-charge constraints of hybrid energy storage systems further increase the difficulty of solving the problem. To address this, after determining the stochastic scenario set, the first optimization objective function, and the first constraint conditions, this application uses the Benders decomposition algorithm based on the stochastic scenario set to process the first optimization objective function and the first constraint conditions, generating the first configuration information of energy storage at the first time granularity.

[0086] In an exemplary embodiment, directly solving a large-scale mixed-integer linear programming problem containing numerous scenarios can be extremely difficult. To avoid this problem, this application, based on a random scenario set, processes the first optimization objective function and the first constraint conditions using the Benders decomposition algorithm. During the process of generating the first configuration information for energy storage at the first time granularity, the random scenario set can be reduced in number using a synchronous back-substitution elimination method or K-means clustering to obtain a representative target scenario set, which can then be assigned a new probability weight π. k This significantly reduces computational complexity while preserving key statistical characteristics of the uncertainty set. Then, based on the target scenario set, the first optimization objective function and the first constraint conditions are processed by the Benders decomposition algorithm to generate the first configuration information of energy storage at the first time granularity.

[0087] In specific application scenarios, when solving for the first configuration information using the Benders decomposition algorithm, the main problem can be set as solving for the first-stage investment decision x; the sub-problems can be set as solving for the optimal operating strategy y for each scenario k, given x. k This generates a Benders cut (a type of linear inequality) and feeds it back to the main problem. This cut contains information about how the operating cost changes with the investment x in this scenario. The main problem and subproblems are solved iteratively until convergence, at which point the main problem... , , This is the optimal first configuration information we are looking for. This method guarantees the lowest total lifecycle cost in an "average sense".

[0088] Step S107: Based on the first configuration information, generate second configuration information for energy storage at the second time granularity.

[0089] Step S108: Based on the second configuration information, generate third configuration information for energy storage at the third time granularity.

[0090] In practical applications, existing energy storage configuration methods are mostly limited to a single time scale, failing to fully consider the manifestations and interactions of random variables at different time scales. For example, most studies focus on daily or hourly optimization; however, renewable energy fluctuations exhibit different characteristics at different time scales such as minute, hour, and day. This single-time-scale optimization cannot effectively coordinate the needs of rapid frequency regulation (second / minute level) and peak shaving (hour level) across multiple time scales. To address this issue, after determining the first configuration information of energy storage at the first time granularity, this application also needs to generate second configuration information of energy storage at the second time granularity based on the first configuration information; and generate third configuration information of energy storage at the third time granularity based on the second configuration information; wherein the first time granularity is greater than the second time granularity, and the second time granularity is greater than the third time granularity, for example, the second time granularity can be day or hour, and the third time granularity can be minute or second, etc. In this way, configuring energy storage according to the first configuration information, the second configuration information, and the third configuration information is equivalent to coordinating the configuration of energy storage from the first time granularity with a large granularity, through the second time granularity with a medium granularity, and finally to the third time granularity with the smallest granularity.

[0091] In an exemplary embodiment, during the process of generating second configuration information for energy storage at a second time granularity based on the first configuration information, an uncertainty vector for processing the energy storage at the second time granularity can be determined. This uncertainty vector can be... This refers to the prediction error or actual value of photovoltaic power output, wind power output, load, and real-time electricity price; historical operating data are collected according to the uncertainty vector to obtain the sample set {ξ1,ξ2,...,ξ}. N Based on the sample set, construct a moment uncertainty set based on one-matrix and two-matrix uncertainties, where the one-matrix represents the mean μ. ^ The two matrices represent the covariance Σ. ^ The set of moments of uncertainty is represented as D = {ℙ∈(Ξ)|E} ℙ [ξ]=μ,E ℙ [(ξ-μ)(ξ-μ) T =Σ,(μ-μ ^ ) T G μ -1 (μ-μ ^ )≤ρ μ ,||Σ-Σ ^ ||F≤ρ Σ}, E ℙ [·] represents the expectation under the distribution ℙ, Ξ represents the support set of the uncertainty vector, and ρ μ and ρ ΣIt is a radius parameter that characterizes the confidence level of moment estimation. The larger the radius, the less trust there is in historical data, and the wider the range of "worst-case" considerations. Based on the first configuration information and the moment uncertainty set, a two-stage sub-Brubon optimization model for energy storage is established at the second time granularity. The two-stage sub-Brubon optimization model is processed by a column and constraint generation algorithm to generate the second configuration information for energy storage at the second time granularity, so as to realize the day-ahead market and intraday scheduling plan with the help of the second configuration information.

[0092] In specific application scenarios, during the process of establishing a two-stage partial Brouban optimization model for energy storage at the second time granularity based on the first configuration information and the moment uncertainty set, the second decision variable for energy storage at the second time granularity can be determined. The second decision variable can include first-stage variables and second-stage variables. The first-stage variable can include the day-ahead decision u, such as the 24-hour planned charging and discharging power p of the energy storage. t sch Contracted electricity volume with the power grid (w) t da These variables must be determined before the uncertainty is realized. The second-stage variables can include real-time balancing operations v(ξ), such as the real-time adjustment power Δp of energy storage. t (ξ), Purchase unbalanced electricity q t (ξ), etc., are adjusted after the uncertainty ξ is realized; based on the moment uncertainty set and the second decision variable, a second optimization objective function for energy storage at the second time granularity is generated, which can be expressed as: , This refers to currently known costs, such as electricity costs under a contract signed according to the plan. Let represent the real-time scheduling cost function, which embodies the minimum cost incurred by performing optimal real-time adjustments after the occurrence of uncertainty ξ. It is an optimization problem itself with parameters u and ξ. The inner max function of the second optimization objective function is used to find the worst-case distribution that maximizes the expected real-time cost. Based on the first configuration information, second constraints for energy storage are generated at the second time granularity. These second constraints may include first-stage and second-stage constraints. The first-stage constraint is that u must satisfy the condition based on x. pow and x cap Given the established day-ahead market bidding rules and energy storage physical constraints, the two-stage constraints are that for each ξ under the distribution ℙ, v(ξ) must satisfy real-time power balance, real-time energy storage operation limits, etc., to ensure that the scheduling plan u has a feasible solution under any possible uncertainty. The second optimization objective function and the second constraint are used as the two-stage sub-Brupbond optimization model of energy storage at the second time granularity.

[0093] In an exemplary embodiment, during the process of generating third configuration information for energy storage at a third time granularity based on the second configuration information, a third decision variable for energy storage at the third time granularity can be determined. The third decision variable may include power allocation instructions, SOC adjustment values, and operating mode switching, etc. Based on the third decision variable, a third optimization objective function for energy storage at the third time granularity is generated. For example, the third optimization objective function could be min α·|Δf|+β·|P. actual -P schedule |+γ·|SOC-SOC ref |, Δf represents the frequency deviation, P actual -P schedule The deviation between actual power and planned power, SOC-SOC ref The deviation of the state of charge from the reference value is represented by α, β, and γ, which are weighting coefficients. Based on the second configuration information, a third constraint condition for energy storage is generated at the second time granularity. The third constraint condition may include frequency response constraint, SOC safety constraint, ramp rate constraint, and circulating current suppression constraint. The frequency response constraint is used to automatically adjust the output power according to the frequency deviation. The SOC safety constraint is used to ensure that the SOC of each energy storage unit is within a safe range. The ramp rate constraint is used to limit the rate of power change to avoid equipment damage. The circulating current suppression constraint is used to prevent excessive circulating current between parallel battery clusters. The third optimization objective function and the third constraint condition are processed by a deep reinforcement learning method to generate the third configuration information for energy storage at the third time granularity. The third configuration information is used to handle real-time power adjustment from the second to the minute level, thereby dealing with high-frequency random interference, such as frequency fluctuations and instantaneous changes in renewable energy output.

[0094] In specific application scenarios, during the process of generating third configuration information for energy storage at the third time granularity by processing the third optimization objective function and the third constraint conditions using deep reinforcement learning methods, the state vector can be determined based on the energy storage frequency, state of charge, power, random disturbances, and periodic pattern learning time. For example, it can be determined based on the normalized frequency deviation Δf. t The frequency-dependent vector in the state vector is determined by the rate of change of frequency d(Δf) / dt, based on the state of charge (SOC) of the lithium battery and supercapacitor. bess_t SOC sc_t The energy storage health status determination state vector is the energy storage's own state vector within the state vector, based on the planned power P from the coordination and scheduling layer. sch_t The deviation between actual and planned power is used to determine the planned and deviation vectors in the state vector, based on the net load fluctuation P predicted in the ultra-short-term forecast. fluct_t Key bus voltage V tThe process involves determining the random disturbance observation vector in the state vector, and the time information vector in the state vector based on the time of day used to learn periodic patterns. Based on the energy storage control method, the action vector is determined. This action vector, relative to the "control commands" output by the deep reinforcement learning agent, is used to determine how to allocate power in response to the state. The action vector can include continuous actions and mixed actions; continuous actions can be 'a'. t =[P bess_ref ,P sc_ref This means directly outputting reference power commands for the lithium battery and supercapacitor to ensure the total power meets constraints. The hybrid action can be a... t =[B,mode], where B is a continuous value representing the proportion of total power demand allocated to the supercapacitor, and mode is a discrete instruction, such as "priority charging," "priority discharging," or "sleep mode," to satisfy power coupling constraints. A reward function is generated based on energy storage evaluation metrics to determine the learning direction of the deep reinforcement learning agent. The reward function can be r... t =-(ω1*Δf t 2 )-(ω2*(P actual -P sch ) 2 )-(ω3*(SOC-0.5) 2 )-(ω4*|ΔP|)+(ω5*R service ), (ω1*Δf t 2 ) represents the frequency stability term, (ω2*(P) actual -P sch ) 2 (ω3*(SOC-0.5)) represents the tracking plan item corresponding to economic efficiency. 2 To avoid overcharging and discharging, there is a SOC maintenance term (ω4*|ΔP|). To reduce equipment wear, there is a smoothing term (ω5*R). service ω represents the incentive corresponding to the revenue from ancillary services. i The weight coefficients for each sub-objective are used to balance safety, economy, and equipment lifespan by adjusting the weights; a deep reinforcement learning agent is determined based on the state vector, action vector, and reward function; the deep reinforcement learning agent processes the third optimization objective function and the third constraint condition to generate third configuration information for energy storage at the third time granularity.

[0095] This application provides an energy storage configuration method, which involves acquiring historical operation data of energy storage; extracting features from the historical operation data according to a first time granularity to obtain a set of random scenarios with coupled random variable statistical characteristics, temporal correlation, and spatial coupling; determining a first decision variable for energy storage at the first time granularity, the first decision variable including energy storage configuration capacity, energy storage configuration power, and energy storage type selection; generating a first optimization objective function for energy storage at the first time granularity based on the first decision variable; generating a first constraint condition for energy storage at the first time granularity; processing the first optimization objective function and the first constraint condition using the Benders decomposition algorithm based on the random scenario set to generate first configuration information for energy storage at the first time granularity; generating second configuration information for energy storage at a second time granularity based on the first configuration information; and generating third configuration information for energy storage at a third time granularity based on the second configuration information; wherein the first time granularity is greater than the second time granularity, and the second time granularity is greater than the third time granularity. In this application, features are extracted from historical operating data at the first time granularity to obtain a set of random scenarios with coupled random variable statistical characteristics, temporal correlation, and spatial coupling. This achieves an accurate characterization of the spatiotemporal correlation characteristics of multivariate random disturbances. Then, based on the first decision variable, the first optimization objective function, and the first constraint, an optimization architecture for energy storage at the first time granularity is constructed. According to the set of random scenarios, the first configuration information of energy storage at the first time granularity is generated through the Benders decomposition algorithm, which significantly improves the adaptability and robustness of the energy storage configuration scheme in real complex environments. Subsequently, a hierarchical planning framework is constructed through the second and third time granularities, enabling the configuration of energy storage according to different time granularities. This optimizes the long-term configuration and short-term operation of energy storage, overcomes the limitations of traditional single-time-scale optimization, and improves the reliability of energy storage configuration.

[0096] To facilitate understanding of the effectiveness of this application's solution, it is assumed that the first time granularity is year or month, the second time granularity is day or hour, and the third time granularity is minute or second. The data processing procedure for this application's energy storage configuration solution is as follows: Figure 2 As shown in Table 1, the results after the experiments and simulations were conducted.

[0097] Table 1. Comparison of this proposed solution with existing solutions.

[0098]

[0099] As shown in Table 1, this application, through a hierarchical planning framework with three time granularities, can simultaneously optimize long-term energy storage configuration and short-term operation, effectively coordinating responses to random disturbances at different time scales such as seconds, minutes, and hours. It also meets various ancillary service requirements such as frequency regulation, peak shaving, and energy time shifting, overcoming the limitations of traditional single-time-scale optimization. Experimental studies show that this application can improve the economic benefits of the energy storage system over its entire lifecycle by 15-25%, while simultaneously improving power supply reliability by 8-12%. Furthermore, by accurately characterizing the spatiotemporal correlation characteristics of multiple random disturbances, this application significantly improves the adaptability and robustness of energy storage configuration schemes in real-world complex environments. Simulation results show that, under the same investment scale, this application can reduce wind and solar curtailment rates by 5-8% and reduce load shortage rates by 3-5%. This application fully considers the technical characteristics and economic differences of different types of energy storage, achieving complementary advantages in power and energy characteristics. Compared with single energy storage configurations, hybrid energy storage systems can improve capacity utilization by 20-30% and extend battery life by 15-20%. Furthermore, this application combines distributed bar optimization and deep reinforcement learning to significantly improve computational efficiency while ensuring solution quality. For example, practical applications show that the configuration scheme provided by this application solves problems 10-20 times faster than traditional methods and is easier to integrate with actual operating systems.

[0100] Please see Figure 3 , Figure 3 This is a schematic diagram of an energy storage configuration system provided in an embodiment of this application.

[0101] An energy storage configuration system provided in this application embodiment may include:

[0102] Historical operation data acquisition module 101 is used to acquire historical operation data of energy storage;

[0103] The random scene set generation module 102 is used to extract features from historical running data according to the first time granularity to obtain a random scene set with coupled random variable statistical features, temporal correlation and spatial coupling.

[0104] The first decision variable determination module 103 is used to determine the first decision variable of energy storage at the first time granularity. The first decision variable includes energy storage configuration capacity, energy storage configuration power and energy storage type selection.

[0105] The first optimization objective function generation module 104 is used to generate a first optimization objective function for energy storage at a first time granularity based on the first decision variable.

[0106] The first constraint generation module 105 is used to generate the first constraint conditions for energy storage at the first time granularity.

[0107] The first configuration information generation module 106 is used to process the first optimization objective function and the first constraint condition based on a random scenario set using the Benders decomposition algorithm to generate the first configuration information of energy storage at the first time granularity.

[0108] The second configuration information generation module 107 is used to generate second configuration information for energy storage at a second time granularity based on the first configuration information.

[0109] The third configuration information generation module 108 is used to generate third configuration information for energy storage at a third time granularity based on the second configuration information.

[0110] Among them, the first time granularity is larger than the second time granularity, and the second time granularity is larger than the third time granularity.

[0111] The energy storage configuration system provided in this application embodiment includes a first decision variable determination module that can be used to: determine a first-stage variable of energy storage at a first time granularity, the first-stage variable including energy storage configuration capacity, energy storage configuration power, and energy storage type selection; determine a second-stage variable of energy storage at a first time granularity, the second-stage variable including moment-to-moment charging power, moment-to-moment discharging power, and moment-to-moment state of charge in random scenarios and concentrated scenarios; and use the first-stage variable and the second-stage variable as the first decision variable.

[0112] This application provides an energy storage configuration system in which a first configuration information generation module can be used to: reduce the number of random scenario sets by synchronous back-substitution elimination or K-means clustering to obtain a target scenario set; and based on the target scenario set, process the first optimization objective function and the first constraint condition by the Benders decomposition algorithm to generate first configuration information for energy storage at the first time granularity.

[0113] This application provides an energy storage configuration system in which a second configuration information generation module can be used to: determine the uncertainty vector for processing energy storage at a second time granularity; collect historical operating data according to the uncertainty vector to obtain a sample set; construct a moment uncertainty set based on one-matrix and two-matrix uncertainties according to the sample set; establish a two-stage sub-Brubern optimization model for energy storage at a second time granularity based on the first configuration information and the moment uncertainty set; and process the two-stage sub-Brubern optimization model through a column and constraint generation algorithm to generate second configuration information for energy storage at a second time granularity.

[0114] This application provides an energy storage configuration system in which a second configuration information generation module can be used to determine a second decision variable for energy storage at a second time granularity; generate a second optimization objective function for energy storage at a second time granularity based on the moment uncertainty set and the second decision variable; generate a second constraint condition for energy storage at a second time granularity based on the first configuration information; and use the second optimization objective function and the second constraint condition as a two-stage sub-Brønsted optimization model for energy storage at a second time granularity.

[0115] This application provides an energy storage configuration system in which a third configuration information generation module can be used to: determine a third decision variable for energy storage at a third time granularity; generate a third optimization objective function for energy storage at a third time granularity based on the third decision variable; generate a third constraint condition for energy storage at a second time granularity based on the second configuration information; and process the third optimization objective function and the third constraint condition using a deep reinforcement learning method to generate third configuration information for energy storage at a third time granularity.

[0116] This application provides an energy storage configuration system in which a third configuration information generation module determines a state vector based on the energy storage frequency, state of charge, power, random disturbances, and periodic pattern learning time; determines an action vector based on the energy storage control method; generates a reward function based on the energy storage evaluation index; determines a deep reinforcement learning agent based on the state vector, action vector, and reward function; and processes the third optimization objective function and third constraints through the deep reinforcement learning agent to generate third configuration information for energy storage at a third time granularity.

[0117] This application also provides an electronic device and a computer-readable storage medium, both of which have the corresponding effects of the energy storage configuration method provided in the embodiments of this application. Please refer to... Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0118] An electronic device provided in this application includes a memory 201 and a processor 202. The memory 201 stores a computer program, and the processor 202 executes the computer program to implement the steps of the energy storage configuration method described in any of the above embodiments.

[0119] Please see Figure 5Another electronic device provided in this application embodiment may further include: an input port 203 connected to the processor 202 for transmitting commands input from the outside to the processor 202; a display unit 204 connected to the processor 202 for displaying the processing results of the processor 202 to the outside; and a communication module 205 connected to the processor 202 for enabling communication between the electronic device and the outside. The display unit 204 may be a display panel, a laser scanning display, etc.; the communication method adopted by the communication module 205 includes, but is not limited to, Mobile High-Definition Link (MHL), Universal Serial Bus (USB), High-Definition Multimedia Interface (HDMI), wireless connection: Wireless Fidelity (WiFi), Bluetooth communication technology, Bluetooth Low Energy communication technology, and communication technology based on IEEE 802.11s.

[0120] This application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the steps of the energy storage configuration method described in any of the above embodiments.

[0121] The computer-readable storage media involved in this application include random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs (compact disc read-only memory), or any other form of storage media known in the art.

[0122] This application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the energy storage configuration method described in any of the above embodiments.

[0123] For descriptions of relevant parts in the energy storage configuration system, electronic device, and computer-readable storage medium provided in this application's embodiments, please refer to the detailed description of the corresponding parts in the energy storage configuration method provided in this application's embodiments; they will not be repeated here. Furthermore, parts of the technical solutions provided in this application that are consistent with the implementation principles of corresponding technical solutions in the prior art have not been described in detail to avoid excessive elaboration.

[0124] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0125] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An energy storage configuration method, characterized in that, include: Obtain historical operational data of energy storage; Feature extraction is performed on the historical operational data according to the first time granularity to obtain a set of random scenarios with coupled random variable statistical features, temporal correlation and spatial coupling; Determine the first decision variable for energy storage at the first time granularity, the first decision variable including energy storage configuration capacity, energy storage configuration power and energy storage type selection; Based on the first decision variable, generate a first optimization objective function for energy storage at the first time granularity; Generate the first constraint condition for energy storage at the first time granularity; Based on the random scene set, the first optimization objective function and the first constraint are processed by the Benders decomposition algorithm to generate first configuration information for energy storage at the first time granularity; Based on the first configuration information, second configuration information for energy storage is generated at a second time granularity. Based on the second configuration information, third configuration information for energy storage at a third time granularity is generated; Wherein, the first time granularity is greater than the second time granularity, and the second time granularity is greater than the third time granularity.

2. The method according to claim 1, characterized in that, The first decision variable for determining energy storage at the first time granularity includes: Determine a one-stage variable for energy storage at the first time granularity, wherein the one-stage variable includes energy storage configuration capacity, energy storage configuration power, and energy storage type selection; Determine the two-stage variables for energy storage at the first time granularity, wherein the two-stage variables include the moment-to-moment charging power, moment-to-moment discharging power, and moment-to-moment state of charge of the random scenario set. The first-stage variables and the second-stage variables are used as the first decision variables.

3. The method according to claim 2, characterized in that, Based on the random scene set, the first optimization objective function and the first constraint are processed using the Benders decomposition algorithm to generate first configuration information for energy storage at the first time granularity, including: The random scene set is reduced in number by synchronous back-substitution elimination or K-means clustering to obtain the target scene set; Based on the target scenario set, the first optimization objective function and the first constraint are processed by the Benders decomposition algorithm to generate first configuration information for energy storage at the first time granularity.

4. The method according to claim 1, characterized in that, The step of generating second configuration information for energy storage at a second time granularity based on the first configuration information includes: Determine the uncertainty vector for processing energy storage at the second time granularity; The historical operational data is collected according to the uncertainty vector to obtain a sample set; Based on the sample set, construct a moment uncertainty set based on one-matrix and two-matrix uncertainties; Based on the first configuration information and the moment uncertainty set, a two-stage sub-Bruvian optimization model for energy storage at the second time granularity is established. The two-stage sub-Brubon optimization model is processed by a column and constraint generation algorithm to generate second configuration information for energy storage at the second time granularity.

5. The method according to claim 4, characterized in that, The step of establishing a two-stage sub-Brønsted optimization model for energy storage at the second time granularity based on the first configuration information and the moment uncertainty set includes: Determine the second decision variable for energy storage at the second time granularity; Based on the moment uncertainty set and the second decision variable, a second optimization objective function for energy storage at the second time granularity is generated; Based on the first configuration information, a second constraint condition for energy storage at the second time granularity is generated; The second optimization objective function and the second constraint are used as a two-stage sub-Brupbond optimization model for energy storage at the second time granularity.

6. The method according to claim 1, characterized in that, The step of generating third configuration information for energy storage at a third time granularity based on the second configuration information includes: Determine the third decision variable for energy storage at the third time granularity; Based on the third decision variable, a third optimization objective function for energy storage is generated at the third time granularity; Based on the second configuration information, a third constraint condition for energy storage is generated at the second time granularity; The third optimization objective function and the third constraint are processed by deep reinforcement learning methods to generate third configuration information for energy storage at the third time granularity.

7. The method according to claim 6, characterized in that, The process of processing the third optimization objective function and the third constraint conditions using deep reinforcement learning methods to generate third configuration information for energy storage at the third time granularity includes: The state vector is determined based on the energy storage frequency, state of charge, power, random disturbances, and periodic pattern learning time. Determine the action vector based on the energy storage control method; A reward function is generated based on the energy storage evaluation indicators; Based on the state vector, the action vector, and the reward function, a deep reinforcement learning agent is determined; The third optimization objective function and the third constraint are processed by a deep reinforcement learning agent to generate third configuration information for energy storage at the third time granularity.

8. An energy storage configuration system, characterized in that, include: The historical operation data acquisition module is used to acquire historical operation data of energy storage. The random scene set generation module is used to extract features from the historical running data according to the first time granularity to obtain a random scene set with coupled random variable statistical features, temporal correlation and spatial coupling. The first decision variable determination module is used to determine the first decision variable of energy storage at the first time granularity. The first decision variable includes energy storage configuration capacity, energy storage configuration power and energy storage type selection. The first optimization objective function generation module is used to generate a first optimization objective function for energy storage at the first time granularity based on the first decision variable. The first constraint generation module is used to generate the first constraint for energy storage at the first time granularity. The first configuration information generation module is used to process the first optimization objective function and the first constraint condition based on the random scene set using the Benders decomposition algorithm to generate first configuration information for energy storage at the first time granularity. The second configuration information generation module is used to generate second configuration information for energy storage at a second time granularity based on the first configuration information. The third configuration information generation module is used to generate third configuration information for energy storage at a third time granularity based on the second configuration information. Wherein, the first time granularity is greater than the second time granularity, and the second time granularity is greater than the third time granularity.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the energy storage configuration method as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the energy storage configuration method as described in any one of claims 1 to 7.

Images